Random Forest: PCA on All Samples Matrix

RE:PCA - I think the goal here would be to calculate a PCA on the entire data-set (i.e. all of the simulations in one big data frame) and for each simulation calculate its particular value for the PCs we'll focus on. I haven't had my coffee yet so apologies if this is convoluted - Each PC will have loadings based on each of the features, so you would then calculate equivalent PC values for the validation/testing set by applying those same loadings, even if they are not technically the PC of the testing set. Does that make sense?

Test Set Evaluation

Below are peformance measures on the test set never touched during model building